Method and apparatus for transferring data between a data bus and a data storage device

Description

[0001] This invention relates generally to error protection of data and more specifically to a method and apparatus for protecting data from errors which occur as data is being transferred between a data bus and associated storage devices.

[0002] A data bus transmits data to and from computing elements and data storage devices using, for error protection, parity bits and check sum symbols. These parity bits and check sum symbols enable the systems to detect all single, double and triple bit transmission error patterns and some patterns with four or more erroneous bits.

[0003] Data on the bus is in the form of data bytes (8-bits), and each data byte has associated with it a parity bit. The data bytes and parity bits are transmitted in multiple-byte packets. The packets contain, in addition to the data bytes and parity bits, an associated check sum symbol. The parity bits and the check sum symbol provide what is commonly referred to as "horizontal" and "vertical" parity, respectively. These terms are associated with a conceptual configuration of a packet as a one-byte wide and "n"-byte long block of data, where n is the number of bytes in the packet.

[0004] Using prior technology, when an interface between the data bus and the storage devices receives a data packet over the bus for storage on the associated storage device, it checks the individual bytes and the entire packet for errors using the parity bits and the check sum symbol, respectively. If there are no detectable errors, the interface converts the bytes to words and buffers the data words in an associated buffer in random access memory (RAM).

[0005] The interface then (i) retrieves from the RAM an appropriate number of data bytes to fill one storage unit on the storage medium, for example, 512 bytes to fill a magnetic disk sector, (ii) encodes the converted data to generate error correction code symbols using an error correction code, and, finally, (iii) sends the data and the error correction symbols as analog signals to the storage device. The interface may also encode the data words and the error correction code symbols to facilitate later demodulation of the stored symbols using, for example, a 2/3 code.

[0006] Errors may be introduced into the data bytes between the time they are found to be error free using the parity bits and the check sum symbol and the time they are encoded using the error correction code. For example, as the error-free data are being transferred to the buffer for pre-encoded storage, they may be sent over one or more faulty connections and arrive at the buffer with one or more errors. The interface encodes this erroneous data to generate corresponding error correction symbols and sends them to the storage medium for recording thereon. When the data and error correction code symbols are later retrieved from the storage medium, the interface uses the error correction symbols to correct the data to the pre-encoded state, that is, to the erroneous state in which they arrived at the buffer. It then sends the erroneous data to a requesting device as "error-free" data. What is needed is a mechanism to protect the data from such errors.

[0007] Prior art patent publication EP-A-0-319,188, teaches a fault tolerant apparatus handling blocks of data with which an error correcting code is associated, while transferring the data to a storage medium. The apparatus uses first and second programmable control devices operable selectively in one of two modes. The first mode monitors generation of an error correction code, and the second mode detects errors in data blocks during transmission. There is no bus interface in said prior art, and no interruption of data transmission if there is an error.

[0008] The invention is a bus interface, which connects to both the data bus and a storage interface, and provides error protection for data from the time it receives data over the data bus until the time it returns the data to the data bus for transmittal to a requesting device. The invention also includes a method of operating the bus interface such that the error protection stays with the data as they travel through the interface to the storage medium and from the storage medium back through the interface to the data bus. If errors are found at any point, the bus interface interrupts the data transfer operation.

[0009] The invention in its broad form resides in a data transmission system as in claim 1 and a method of transmitting data as recited in claim 9.

[0010] The method is summarized as follows. When the bus interface receives a data packet from the data bus it determines if the packet and the individual data bytes therein are free of errors detectable by means of the received parity bits and check sum symbol. If the interface does not detect any errors in the received data bytes it converts the data bytes to two-byte data words and associates with each word the two received parity bits formerly associated with the data bytes. Next, using an error detection code (EDC), the interface encodes the data words in groups which are a predetermined number of words long, and generates for each group a number of EDC symbols. The bus interface retains the associated parity bits, but does not encode them.

[0011] After the encoding, the interface again checks each of the encoded data words for errors using the retained parity bits. If the data words contain errors, the bus interface interrupts the data transfer operation. However, if the data words are still error-free, the interface sends the various groups and the associated EDC symbols to a series of "linked" buffers, with each group being stored in a single buffer. The buffers are linked by pointers, each of which points from a buffer associated with earlier received data to a next buffer associated with later received data. Each of the buffers holds the amount of data required for one section of the storage medium, for example, the data for one disk sector, where the data consists of data words and associated EDC symbols.

[0012] After the packet data are stored in the linked buffers, a microprocessor retrieves the EDC symbols from each of the buffers associated with the packet and encodes them by combining them with the address of the disk sector which is designated to store the associated data. Thereafter, the storage interface encodes the data, that is, the data words and the encoded EDC symbols, using an error correction code (ECC), and generates a predetermined number of ECC symbols. It then sends the data and the ECC symbols to the storage device.

[0013] When the data and ECC symbols are later retrieved from the storage device and returned to the storage interface, the interface uses the retrieved ECC symbols to correct the data, as necessary. It then sends the data to the buffers. Next, the microprocessor retrieves the EDC symbols from the buffers, combines them with the address of the designated sector and returns them to the buffer. If the storage device retrieved the data from the wrong sector, the combining of the designated sector address with the retrieved EDC symbols introduces into the EDC symbols one or more errors which will be detected by the bus interface.

[0014] The bus interface then retrieves the data from the buffer and reverses the error protection routine it went through to prepare the data for storage. The bus interface thus (i) generates parity bits for each of the retrieved data words, (ii) uses the retrieved EDC symbols to determine if there are errors in the data words, (iii) converts the data words to data bytes and associates one of the generated parity bits with each of the bytes, and (iv) generates a check sum symbol for the packet. If the bus interface detects an error at any point it stops the data transfer operation and interrupts the microprocessor.

[0015] If the bus interface does not detect any errors, it sends over the bus the data bytes, in the form of a packet. While it is sending the packet, the bus interface performs one more error detection operation to ensure that the connections between the interface and the bus have not introduced errors into the data. The bus interface thus reads back the data it sends, and compares them with the corresponding data bytes and parity bits it tried to send. If everything matches, the transmission is error-free. Otherwise, the bus interface interrupts the data transfer operation.

[0016] The above and further advantages of the invention may be better understood by referring to the following description in conjunction with the accompanying drawings, in which:

Figure 1 is a block diagram of a data bus, an associated data storage device and the interfaces between them;

Figure 2 is a block diagram of a bus interface shown in Figure 1;

Figure 3A is a flow chart of the operations of the bus interface in transferring data from the bus to a storage interface;

Figure 3B is a flow chart of the operations of the storage interface in transferring data to the storage device;

Figure 4A is a flow chart of the operations of the storage interface in transferring data from the storage device to the bus interface;

Figure 4B is a flow chart of the operations of the bus interface in transferring data to the data bus; and

Figure 5 is a block diagram of "linked" data buffers.

[0017] Figure 1 depicts, in block diagram form, a data bus,9, a mass storage device 18 and the interfaces 10 and 14 between them. The bus interface 10 and the storage interface 14 facilitate the transfer of data,between the bus and the storage device 18, under the control of a microprocessor 12.

[0018] The bus interface 10 receives data-from the bus and transmits data to the bus in the form of multi-byte packets. Each packet also contains a number of parity bits, one for each data byte, and a check sum symbol for error protection, as well as a packet header which includes packet specific information such as the source and destination of the packet, the length of the packet, and so forth.

[0019] When the bus interface 10 receives a packet from the bus for storage on the associated storage device 18, it checks the data bytes for errors using the parity-bits and check sum symbol by means of conventional techniques. While it is checking the packet for errors using the check sum, it also converts the data bytes to two-byte words and encodes a predetermined number of them, using an error detection code EDC, to generate EDC symbols. It then checks the words which it encoded for errors using the two associated parity bits received on the data bus 9. Next, it sends the data words and associated EDC symbols to one or more buffers in a series of "linked" buffers in RAM 16. Thus the bus interface 10, under the control of the microprocessor 12, sends the data through the storage interface 14 to designated buffer(s) in the RAM 16. Each of the linked buffers holds one storage unit of data, for example, one sector of data, where the data includes both the data words and the EDC symbols. The operation of the linked buffers is discussed in more detail with reference to Figure 5 below.

[0020] When the bus interface 10 completes the transfer of data to the RAM 16, it sends an interrupt to the microprocessor 12. The microprocessor 12 then retrieves the EDC symbols from the buffers and encodes the symbols by combining them with the addresses of the various sectors on the storage device which are designated to store the data. The microprocessor 12 thus exclusive-ORs the EDC symbols in the first buffer associated with the packet with the address symbols associated with the first sector designated to store the data and returns the resulting symbols to the appropriate buffer. It then exclusive-ORs the EDC symbols from the next buffer with the address symbols associated with a next sector and returns the resulting symbols to the buffer, and so forth.

[0021] When the storage device 18 is ready for the data, the storage interface 14 retrieves the data from RAM 16, encodes it using an error correction code (ECC), and generates corresponding ECC symbols. Next, it transfers the data words, encoded EDC symbols and ECC symbols to the storage device 18. The storage device 18, using write heads (not shown), then records the data and associated symbols in the appropriate sector, or section, of the storage medium.

[0022] The bus interface 10 provides error protection for data from the time it receives the data from the data bus. This error protection follows the data to the storage medium and back to the bus interface 10. Thus it protects the data from errors which are caused by faulty connections, wires, etc. in the storage interface 12, the associated RAM 16 and/or connections between them. This error protection, together with the ECC protection provided by the storage interface 14, ensures that only error-free data from the appropriate sector are transferred to the data bus 9. The bus interface 10 detects these types of errors in data retrieved from storage by reversing the operations set forth above, as discussed in more detail below with reference to Figures 4A and 4B.

[0023] Figure 2 depicts the bus interface 10 in more detail. The bus interface 10 includes in a receive path a receive/transmit circuit 20, a byte parity-checker 22, a check sum circuit 24, a byte/word converter 26, an EDC encoder/decoder 28, and a word parity-checker 30. With reference also to Figures 3A and 3B, when the receive/transmit circuit 20 receives a data packet over the data bus it sends each data byte and associated parity bit to byte parity-checker 22 as it receives them. Byte parity-checker 22 uses the received parity bit to check the data byte for errors (Steps 100-102). If the byte parity checker 22 determines that the parity is not correct for a given byte, it asserts an "external parity error" (EPE) signal, which interrupts the microprocessor 12 and stops the data transfer operation. The bus interface 10 also stores in an internal status register 11 an indication that it detected a data byte parity error (Steps 106-107). The microprocessor 12 later retrieves the information in register 11 when it responds to the interrupt signal.

[0024] If the parity of a given data byte is correct, the parity checker 22 sends the data byte to the check sum circuit 24 and the associated parity bit to byte/word converter 26. The check sum circuit 24 includes the data byte in its check sum calculation and then sends the byte to the byte/word converter 26 (Step 104). When the converter 26 receives two data bytes, it converts the bytes to a word, sends the two parity bits associated with that word to the word parity-checker 30, and sends the word to EDC encoder/decoder 28 (Steps 108-110). As is understood by those skilled in the art, the converter 26 converts the data bytes to the format required by the storage devices 18. If, for example, the devices 18 required 32-bit words, the converter 26 would convert four data bytes to a 32-bit word and associate with the word the four parity bits associated with the individual bytes.

[0025] The EDC encoder/decoder 28 includes the word in its EDC symbol generation and sends the word to the data word parity checker 30. The word parity-checker 30 then uses the parity bits it received from the converter 26 to check the word it received from the encoder/decoder 28 for errors. If it does not detect any errors, it sends the word to a designated buffer in RAM 16 (Steps 112, 114-117). When a sufficient number of data words have passed through the EDC encoder/decoder 28, the encoder/decoder sends to the particular RAM buffer which is holding the data words a number of associated EDC symbols (Steps 116, 122). If the word parity-checker 30 finds an error in any data word it asserts an "internal parity error" (IPE) signal, which interrupts both the microprocessor 12 and the data transfer operation. The bus interface 10 then stores information relating to the IPE in its status register 11 (Steps 120-121).

[0026] As the bus interface 10 receives more of the packet, it generates additional EDC symbols and fills additional buffers in RAM 16. Once a complete packet is received, the check sum circuit 24 determines if the vertical parity is correct (Steps 124-126). If it is, the check sum circuitry 24 leaves de-asserted the EPE signal. Otherwise, it asserts the EPE signal, which interrupts the microprocessor 12 and stops the data transfer operation, and stores the appropriate information in its status register 11 (Steps 128-129).

[0027] If the packet does not contain enough data to fill all of the buffers to which the data are sent, the bus interface 10 zero-fills the remaining locations in the last buffer associated with the packet (steps 118-119). The bus interface 10 thus generates an appropriate number of all zero data bytes and associated parity bits, converts the bytes to words and includes them in the EDC. It then sends the data and associated EDC symbols to the buffer. In this way, the bus interface protects the buffered data from errors which change all-zero filler symbols to non-zero symbols which may be mistaken as data. The bus interface thus sends to the buffers a number of complete sectors of data, and eliminates the need for the storage interface 14 or microprocessor 12 to generate zero-fill symbols in order to fill completely each designated sector of the storage medium. The storage device 18 must receive a full sector of data from the storage interface 14, because it can only "write" data to the storage medium in complete sector units.

[0028] When the entire packet has been sent to RAM 16, the bus interface 10 sends an interrupt to the microprocessor 12 (Step 130). The microprocessor 12 thereafter retrieves from the first buffer associated with the packet the EDC symbols and encodes them by combining them with the address of the storage medium sector designated to store the data and returns the encoded EDC symbols to the buffer (Steps 132-134). The microprocessor 12 then repeats the encoding for each buffer, using appropriate sector addresses (Steps 136-137).

[0029] Thereafter, when the storage device 18 is ready to record the data, the storage interface 14 retrieves the data from the buffers, encodes it using an ECC and generates ECC symbols (Step 138). Finally, it sends the data and associated symbols to the storage device 18 (Step 139). The storage interface 14 may also encode the data to facilitate later demodulation using, for example, a 2/3 code, as appropriate.

[0030] Referring now also to Figures 4A and 4B, when data is retrieved from storage, the storage interface 14 decodes the data using the retrieved ECC symbols and accumulates in RAM 16 a sector of data (steps 140-148). The data includes data words and EDC symbols, and the ECC syndromes symbols generated by the decoding. Using the buffered ECC syndrome symbols and conventional error correction mechanisms (not shown), the storage interface 14 and/or the microprocessor 12 corrects errors in the retrieved data (Step 142). If all the errors cannot be corrected, the microprocessor 12 stops the data transfer operation and indicates to a system controller that an error has been detected (Steps 144-146). The controller can then either try to re-read the sector or cancel the data transfer operation all together, as appropriate.

[0031] If all the detectable errors in the data are corrected, the microprocessor 12 retrieves from the buffer the EDC symbols, exclusive-OR's them with the address of the sector from which the storage device 18 was supposed to have retrieved the data and returns the resulting symbols to the buffer (Step 150). It thus introduces errors into the EDC symbols if the data was read from the wrong sector, or if the data was previously written to the wrong sector, that is, the sector contains data intended for a different sector, and prevents the otherwise error-free, but still incorrect, data from being sent to the data requesting device.

[0032] The bus interface 10 retrieves the data from the buffer and generates, for each data word, two parity bits in word parity-generator 29 (Step 152). Next, it sends the data word to EDC encoder/decoder 28 and the parity bits to byte/word converter 26. The EDC encoder/decoder 28 includes the data word in its EDC decoding and sends the word to the converter 26 (Steps 154).

[0033] The byte/word converter 26 converts the data word to two data bytes and associates with each byte one of the generated parity bits (Step 156). It then sends the bytes to check sum circuit 24 and the parity bits to the receive/transmit circuit 20. The check sum circuit 24 includes the data bytes in its check sum calculations and sends the bytes to the transmit circuit 20 to join their parity bits (Step 158).

[0034] After a full buffer of received data words and EDC symbols have been sent through the EDC encoder/decoder 28, the encoder/decoder uses the retrieved EDC symbols to determine if the data is still error-free (Steps 160-164). If they are, the encoder/decoder leaves de-asserted an "EDC error" signal. Otherwise, it asserts the EDC error signal, which stops the transmission operation and interrupts the microprocessor 12. The bus interface 10 then stores information in its internal status register 11 which indicates that an EDC error was detected (Steps 166-167).

[0035] Once all the data bytes and associated parity bits requested by the requesting device have passed through the check sum circuit 24, the circuit produces a check sum symbol and sends the symbol to the receive/transmit circuit 20 for transmission (Steps 168-170). If the requesting device requests data which does not correspond to one or more complete sectors, the bus interface 10 includes only the requested data in its check sum calculation. However, it includes all the data in its EDC calculation in order to ensure that the requested data symbols are error-free. Thus the bus interface 10 may send the requested data to the transmit circuit 20 and then keep the circuit waiting for an indication of a valid EDC decode operation.

[0036] The receive/transmit circuit 20 either at this point, or earlier in parallel with the various on-going error protection operations, transmits the requested data bytes, parity bits and check sum symbol onto the data bus in byte-serial form. At the same time, actually after a short delay which corresponds to the propagation delay through the bus driving circuitry (not shown), the receive/transmit circuit 20 reads back the information it transmitted onto the bus. It then compares the read-back information with the information it tried to send (Steps 170-174). If everything matches, the circuit 20 continues its transmission operation (Steps 176, 180). Otherwise, the circuit 20 asserts a transmission-error signal, which interrupts the microproccessor 12 and stops the transmission operation. The bus interface 10 then stores in its status register information which indicates that it detected a transmission error (Steps 178-179). The microprocessor 12 can then re-read the appropriate sector and/or RAM buffers, attempt to again send the information on the bus or notify the requesting device that the data transfer can not be made, as appropriate.

[0037] Referring now to Figure 5, the RAM 16 stores data in a linked list of buffers. The data which are first received are stored in a first buffer B₁. When that buffer is full the next data are stored in a next buffer B₂, and so forth. The first buffer B₁ contains, along with the data, a pointer p₁ which points to the next buffer B₂. Buffer B₂ contains a pointer p₂ which points to the next buffer B₃, and so forth. Thus if the data from one packet are stored in "s" buffers, each of the buffers, except the last buffer, B_s, points to the next buffer in which the packet data are stored. If another packet has been received, the last buffer B_s contains a pointer to the first buffer associated with this next packet. Accordingly, each of the buffers has a pointer which points to a next buffer associated with data later received over the bus. These buffers are thus "linked" together. If all the received data are stored, the last buffer contains either a "0" pointer, which indicates the end of a list, or a pointer to an un-used buffer, which indicates where the next received data will be stored. The pointers essentially order the buffers, such that the data are sent to a requesting device or to storage in the same order they were received by the RAM 16.

[0038] Each buffer B_i includes, in addition to the data and a pointer p_i, one or more synchronization characters 34, and a header and associated header EDC symbols 36-38. The synchronization characters 34, which are written to the buffer at the same time that the data are written, signify a valid data buffer. They are patterns of binary ONES and ZEROS which enable the bus interface 10 to detect errors such as "stuckats", where one wire is held high or low, or two-bit "bridges", which are connections between two bits that force them to the same value. For example, the characters may be (in hexadecimal notation) AAAA5555H, and the bus interface examines the corresponding bits in each word which are "0"-"1" pairs to detect the aforementioned errors.

[0039] The header 36 in the first buffer associated with a particular packet contains packet related information such as source and destination addresses, length of packet, and so forth. The header is protected by header EDC symbols 38. If the bus interface 10 detects errors in the packet header, it asserts an error signal which interrupts the microprocessor 12 and ends the data transfer.

[0040] When a data transfer operation is complete, that is, an entire packet has been transferred from the bus interface 10 to the RAM 16, the bus interface over writes one of the synchronization characters in the first of the linked buffers associated with the transferred packet. The bus interface 10 thus writes information concerning the status of the transfer, for example, the number of bytes transferred and so forth, which is later used by the microprocessor 12 in assessing the transfer operation.

[0041] As set forth above, the bus interface 10 protects the data from the time it receives it from the data bus until it sends the data, in the appropriate format, through the storage interface 14 to RAM 16. Accordingly, errors introduced by bad connections, stuck bits, etc. in either the bus interface 10, the storage interface 14 or connections between them and/or the storage device 18 are detected either when the data is being converted to a format used by the storage devices or when the data is later retrieved and converted back to the format required by the bus. Thus the bus interface prevents erroneous data from being sent to a requesting device.

[0042] The bus interface 10 performs both write and retrieval operations. If errors are detected in a write operation, retrieval operations may continue, as long as the mechanism introducing errors into the write data does not also introduce errors into the retrieved data. Similarly, write operations may continue when errors are detected in retrieved data. A system controller may keep track of where errors are detected in a data transfer operation, in order to determine what is causing the errors. If, for example, the controller determines that data transmitted by the bus interface often includes errors which are discovered during the read-back phase of the transmission, it can test the data bus drivers and determine if one or more of them are faulty.

[0043] The foregoing description has been limited to a specific embodiment of this invention. It will be apparent, however, that variations and modifications may be made to the invention, with the attainment of some or all of its advantages which fall within the scope of the appended claims.

Claims

1. A data transmission system for transmitting data between a data bus (9), which transmits data in the form of multi-byte packets that include a plurality of data bytes and associated parity bits and a packet check sum symbol, and a data storage device (18), the system comprising:

A. a microprocessor (12) for controlling a data transfer operation;

B. a bus interface (10) which includes --

i. receiving means (20) for receiving data from the data bus,

ii. byte parity check means (22) for detecting errors in received data bytes using received parity bits,

iii. converter means (26) for converting the data bytes to data words and associating with each data word a predetermined number of parity bits,

iv. word parity check means (30) for detecting errors in the data words using the associated parity bits,

v. encoder/decoder means (28) for encoding a predetermined number of data words and generating error detection symbols,

vi. check sum error detecting means (24) for detecting errors in a received multi-byte packet using the received packet check sum symbol,

-- the bus interface (10) including means for asserting an error signal, interrupting the microprocessor (12) and stopping a data transmission operation whenever it detects an error;

C. buffer means (16) for receiving from the bus interface data in groups containing a predetermined number of data words and associated error detection symbols and storing each group in a designated buffer;

D. a storage interface (14) for retrieving the data groups one at a time from the buffer means (16) and encoding the retrieved data, using an error correction code, to generate error correction symbols; and

E. a storage device (18) for receiving the data and error correction symbols from the storage interface (14) and storing the data and the error correction symbols in a storage unit on a storage medium.

2. The data transmission system of claim 1, wherein the buffer means (16) includes a plurality of buffers (B₁, B₂) which are the size of a data storage unit associated with the storage media, the buffers being linked by pointers (P_i) which point the one buffer to a next buffer, in the order in which the buffers are filled.

3. The data transmission system of claim 2, wherein the encoder/decoder means (28) encodes the number of data words required for a data storage unit to generate the error detection symbols.

4. The data transmission system of claim 3, wherein the bus interface (10) further includes means for generating all-zero bytes to zero-fill a buffer (B₁, B₂) if the received data does not completely fill each of the buffers associated with the received packet, the bus interface generating associated parity bits and including the all-zero bytes in error detection symbol calculations.

5. The data transmission system of claim 4, wherein the microprocessor (12) encodes the error detection symbols in a buffer by combining the error detection symbols with symbols associated with the address of a data storage unit designated to hold the data, before the error detection symbols are encoded using the error correction code.

6. The data transmission system of claim 5, wherein the system further includes:

A. in the buffer means (16), means for receiving data from the storage interface and storing the data in storage-unit sized groups in the buffers (B₁, B₂);

B. in the bus interface (10) --

i. in the word parity check means (30), generating means (29) for generating parity bits for a data word retrieved from the buffer means (16),

ii. in the converter means (26), means for converting received data words to data bytes and associating with each data byte a parity bit generated by the word parity check, and

iii. transmitting means (20) for transmitting data on to the data bus in the form of multi-byte packets;

-- the bus interface including means for checking the retrieved data words for errors using the retrieved error detection symbols before transmitting the data in the form of a multi-byte packet on to the data bus.

7. The data transmission system of claim 6, wherein the microprocessor (12) removes from the error detection symbols in a buffer the address of the designated storage section associated with the data in the buffer before making the error detection symbols and the associated data words available to the bus interface.

8. The data transmission system of claim 7, wherein the bus interface (10) also includes data read-back means for:

A. reading back data transmitted on the data bus by the transmitting means, and

B. comparing the read back data to the data transmitted by the transmitting means to determine if the transmitted data contains errors.

9. A method of transmitting data between a data bus (9) and data storage devices (18), the data transmitting method comprising the steps of:

A. receiving from the data bus (9) a data packet which includes a plurality of data bytes and associated parity bits and a packet check sum symbol;

B. detecting (100) errors in the received data bytes using the received parity bits and asserting (106) an error signal and interrupting a data transfer operation if any errors are detected;

C. converting (108) the data bytes to data words and associating with each data word a predetermined number of received parity bits, if no errors are detected in the data bytes;

D. encoding (110) a predetermined number of data words using an error detection code and generating error detection symbols;

E. detecting (112) errors in the data words using the associated parity bits and asserting (120) an error signal and interrupting the data transfer operation if any errors are detected;

F. storing (117) the predetermined number of data words and associated error detection symbols in a buffer (16; B₁, B₂) sized to hold one storage unit of data and linking this buffer by pointers (P_i) with buffers associated with data from the same data packet or a later received packet, if no errors are detected in the predetermined number of data words;

G. repeating steps D-F;

H. when all the data bytes in a data packet have been encoded and stored, detecting (126) errors in the data packet using the packet check sum symbol and asserting (128) an error signal and interrupting the data transfer operation if any errors are detected;

I. encoding (134) the data words and the associated error detection symbols in a first buffer (B₁) associated with the data packet using an error correction code and generating error correction symbols;

J. storing (139) the data words, associated error detection symbols and the error correction symbols in a storage unit of an associated storage device (18); and

K. repeating steps I and J until all the buffered data words have been stored in storage units of the associated storage device.

10. The method of transmitting data of claim 9, wherein the method further comprises the steps of:

L. retrieving (140) from a storage unit of the associated data storage device (18), the stored data words and associated error detection and error correction symbols;

M. detecting and correcting (142) errors in the data words and error detection symbols using the error correction symbols;

N. asserting (146) an error signal and interrupting the data transfer operation if all detected errors cannot be corrected;

O. generating (152) for each data word a predetermined number of parity bits, if all detected errors are corrected;

P. detecting (154) errors in the data words using the error detection symbols and asserting an error signal and interrupting the data transfer operation if any errors are detected;

Q. converting (156) the data words to data bytes and associating with each data byte one of the parity bits associated with the corresponding data word, if no errors are detected;

R. repeating steps L-Q until the data associated with a data packet has been retrieved;

S. generating a check sum symbol for the data packet; and

T. transmitting (170) the data packet on the data bus.

11. The method of transmitting data of claim 10, wherein the method further includes the step (134) of combining the address of the storage medium unit designated to store the data with the error detection symbols associated with the data before encoding (134) symbols using the error correction code.

12. The method of transmitting data of claim 11, wherein the method still further includes the step (150) of removing the address of the storage medium unit designated to store the data from the error detection symbols associated with data retrieved from the storage medium after decoding the symbols using the error correction code but before using the error detection symbols to detect errors in the retrieved data.

Ansprüche

1. Datenübertragungssystem zum Übertragen von Daten zwischen einem Datenbus (9), der Daten in der Form von Mehr-Byte-Paketen überträgt, die eine Vielzahl von Datenbytes und zugeordneten Paritätsbits und ein Paketprüfsummensymbol umfassen, und einer Datenspeichervorrichtung (18), wobei das System umfaßt:

A. einen Mikroprozessor (12) zum Steuern einer Datenübertragungsoperation,

B. eine Busschnittstelle (10), die aufweist --

i. eine Empfangseinrichtung (20) zum Empfangen von Daten vom Datenbus,

ii. eine Byteparitätsprüfeinrichtung (22) zum Erfassen von Fehlern in empfangenen Datenbytes mittels empfangener Paritätsbits,

iii. eine Umsetzereinrichtung (26) zum Umsetzen der Datenbytes in Datenwörter und zum Zuordnen einer vorbestimmten Anzahl von Paritätsbits zu jedem Datenwort,

iv. eine Wortparitätsprüfeinrichtung (30) zum Erfassen von Fehlern in den Datenwörtern mittels der zugeordneten Paritätsbits,

v. eine Codier/Decodiereinrichtung (28) zum Codieren einer vorbestimmten Anzahl von Datenwörtern und zum Erzeugen von Fehlererfassungssymbolen,

vi. eine Prüfsummenfehlerdetektoreinrichtung (24) zum Erfassen von Fehlern in einem empfangenen Mehr-Byte-Paket mittels des empfangenen Paketprüfsummensymbols,

-- wobei die Busschnittstelle (10) eine Einrichtung aufweist, um ein Fehlersignal zu sichern, den Mikroprozessor (12) zu unterbrechen und eine Datenübertragungsoperation zu stoppen, wann immer sie einen Fehler erfaßt,

C. eine Puffereinrichtung (16), um von der Busschnittstelle Daten in Gruppen zu empfangen, die eine vorbestimmte Anzahl von Datenwörtern und zugeordnete Fehlererfassungssymbole enthalten, und um jede Gruppe in einem bezeichneten Puffer zu speichern,

D. eine Speicherschnittstelle (14) zum Wiedergewinnen der Datengruppen, eine zu einer Zeit, von der Puffereinrichtung (16) und zum Codieren der wiedergewonnenen Daten mittels eines Fehlerkorrekturcodes, um Fehlerkorrektursymbole zu erzeugen, und

E. eine Speichervorrichtung (18) zum Empfangen der Daten und Fehlerkorrektursymbole von der Speicherschnittstelle (14) und zum Speichern der Daten und Fehlerkorrektursymbole in einer Speichereinheit auf einem Speichermedium.

2. Datenübertragungssystem nach Anspruch 1, bei dem die Puffereinrichtung (16) eine Vielzahl von Puffern (B₁, B₂) umfaßt, die die Größe einer Datenspeichereinheit haben, die den Speichermedien zugeordnet ist, wobei die Puffer durch Hinweismarken (P_i), die von einem Puffer zum nächsten Puffer verweisen, in der Reihenfolge verkettet sind, in der die Puffer gefüllt werden.

3. Datenübertragungssystem nach Anspruch 2, bei dem die Codier/Decodiereinrichtung (28) die Anzahl von Datenwörtern codiert, die für eine Datenspeichereinheit erforderlich sind, um die Fehlererfassungssymbole zu erzeugen.

4. Datenübertragungssystem nach Anspruch 3, bei dem die Busschnittstelle (10) weiterhin eine Einrichtung zum Erzeugen von All-Null-Bytes zum Null-Füllen eines Puffers (B₁, B₂) aufweist, wenn die empfangenen Daten nicht vollständig jeden der mit dem empfangenen Paket zugeordneten Puffer füllen, wobei die Busschnittstelle zugeordnete Paritätsbits erzeugt und die All-Null-Bytes in Fehlererfassungssymbolberechnungen einschließt.

5. Datenübertragungssystem nach Anspruch 4, bei dem der Mikroprozessor (12) die Fehlererfassungssymbole in einem Puffer codiert, indem er die Fehlererfassungssymbole mit Symbolen kombiniert, die der Adresse einer Datenspeichereinheit zugeordnet sind, die bezeichnet ist, um die Daten zu halten, bevor die Fehlererfassungssymbole mittels des Fehlerkorrektorcodes codiert werden.

6. Datenübertragungssystem nach Anspruch 5, bei dem das System weiterhin umfaßt:

A. in der Puffereinrichtung (16) eine Einrichtung zum Empfangen von Daten von der Speicherschnittstelle und zum Speichern der Daten in nach Speichereinheiten bemessenen Gruppen in den Puffern (B₁, B₂),

B. in der Busschnittstelle (10) --

i. in der Wortparitätsprüfeinrichtung (30) eine Erzeugungseinrichtung (29) zum Erzeugen von Paritätsbits für ein von der Puffereinrichtung (16) wiedergewonnenes Datenwort,

ii. in der Umsetzereinrichtung (26) eine Einrichtung zum Umsetzen der empfangenen Datenwörter in Datenbytes und zum Zuordnen eines durch die Wortparitätsprüfung erzeugten Paritätsbits zu jedem Datenbyte, und

iii. eine Übertragungseinrichtung (20) zum Übertragen von Daten auf den Datenbus in der Form von Mehr-Byte-Paketen,

-- wobei die Busschnittstelle eine Einrichtung aufweist, um die wiedergewonnenen Datenwörter auf Fehler mittels der wiedergewonnenen Fehlererfassungssymbole zu prüfen, bevor die Daten in der Form eines Mehr-Byte-Paketes auf den Datenbus übertragen werden.

7. Datenübertragungssystem nach Anspruch 6, bei dem der Mikroprozessor (12) aus den Fehlererfassungssymbolen in einem Puffer die Adresse des bezeichneten Speicherabschnittes entfernt, der Daten in dem Puffer zugeordnet ist, bevor die Fehlererfassungssymbole und die zugeordneten Datenwörter für die Busschnittstelle verfügbar gemacht sind.

8. Datenübertragungssystem nach Anspruch 7, bei dem die Busschnittstelle (10) außerdem eine Datenrückleseeinrichtung aufweist, um:

A. auf dem Datenbus durch die Übertragungseinrichtung übertragene Daten rückzulesen, und

B. die rückgelesenen Daten mit den durch die Übertragungseinrichtung übertragenen Daten zu vergleichen, um zu bestimmen, ob die übertragenen Daten Fehler enthalten.

9. Verfahren zum Übertragen von Daten zwischen einem Datenbus (9) und Datenspeichervorrichtung (18), wobei das Datenübertragungsverfahren die folgenden Schritte umfaßt:

A. Empfangen eines Datenpaketes von Datenbus (9), das eine Vielzahl von Datenbytes und zugeordnete Paritätsbits und ein Paketprüfsummensymbol enthält,

B. Erfassen (100) von Fehlern in den empfangenen Datenbytes mittels der empfangenen Paritätsbits und Sichern (106) eines Fehlersignales und Unterbrechen einer Datenübertragungsopera tion, wenn Fehler erfaßt werden,

C. Umsetzen (108) der Datenbytes in Datenwörter und Zuordnen einer vorbestimmten Anzahl von empfangenen Paritätsbits zu jedem Datenwort, wenn keine Fehler in den Datenbytes erfaßt werden,

D. Codieren (110) einer vorbestimmten Anzahl von Datenwörtern mittels eines Fehlererfassungscodes und Erzeugen von Fehlererfassungssymbolen,

E. Erfassen (112) von Fehlern in den Datenwörtern mittels der zugeordneten Paritätsbits und Sichern (120) eines Fehlersignales und Unterbrechen der Datenübertragungsoperation, wenn irgendwelche Fehler erfaßt werden,

F. Speichern (117) der vorbestimmten Anzahl von Datenwörtern und der zugeordneten Fehlererfassungssymbole in einem Puffer (16, B₁, B₂), das bemessen ist, um eine Speichereinheit von Daten zu halten, und Verketten dieses Puffers durch Hinweismarken (P_i) mit Puffern, die Daten von dem gleichen Datenpaket oder einem später empfangenen Paket zugeordnet sind, wenn keine Fehler in der vorbestimmten Anzahl von Datenwörtern erfaßt werden,

G. Wiederholen der Schritte D bis F,

H. wenn alle Datenbytes in einem Datenpaket codiert und gespeichert sind, Erfassen (126) von Fehlern in dem Datenpaket mittels des Paketprüfsummensymboles und Sichern (128) eines Fehlersignales und Unterbrechen der Datenübertragungsoperation, wenn irgendwelche Fehler erfaßt werden,

I. Codieren (134) der Datenwörter und der zugeordneten Fehlererfassungssymbole in einem ersten Puffer (B₁), das dem Datenpaket zugeordnet ist, mittels eines Fehlerkorrekturcodes und Erzeugen von Fehlerkorrektursymbolen,

J. Speichern (139) der Datenwörter, zugeordneter Fehlererfassungssymbole und der Fehlerkorrektursymbole in einer Speichereinheit einer zugeordneten Speichervorrichtung (18) und

K. Wiederholen der Schritte I und J, bis alle gepufferten Datenwörter in Speichereinheiten der zugeordneten Speichervorrichtung gespeichert sind.

10. Verfahren zum Übertragen von Daten nach Anspruch 9, bei dem das Verfahren weiterhin die folgenden Schritte aufweist:

L. Wiedergewinnen (140) der gespeicherten Datenwörter und der zugeordneten Fehlererfassungsund Fehlerkorrektursymbole aus einer Speichereinheit der zugeordneten Datenspeichervorrichtung (18),

M. Erfassen und Korrigieren (142) von Fehlern in den Datenwörtern und Fehlererfassungssymbolen mittels der Fehlerkorrektursymbole,

N. Sichern (146) eines Fehlersignales und Unterbrechen der Datenübertragungsoperation, wenn alle erfaßten Fehler nicht korrigiert werden können,

O. Erzeugen (152) einer vorbestimmten Anzahl von Paritätsbits für jedes Datenwort, wenn alle erfaßten Fehler korrigiert sind,

P. Erfassen (154) von Fehlern in den Datenwörtern mittels der Fehlererfassungssymbole und Sichern eines Fehlersignales und Unterbrechen der Datenübertragungsoperation, wenn irgendwelche Fehler erfaßt sind,

Q. Umsetzen (156) der Datenwörter in Datenbytes und Zuordnen eines der Paritätsbits, das dem entsprechenden Datenwort zugeordnet ist, zu jedem Datenbyte, wenn keine Fehler erfaßt werden,

R. Wiederholen der -Schritte L bis Q, bis die einem Datenpaket zugeordneten Daten wiedergewonnen sind,

S. Erzeugen eines Prüfsummensymboles für das Datenpaket, und

T. Übertragen (170) des Datenpaketes auf dem Datenbus.

11. Verfahren zum Übertragen von Daten nach Anspruch 10, bei dem das Verfahren weiterhin den Schritt (134) des Kombinierens der Adresse der Speichermediumeinheit, die bezeichnet ist, um die Daten zu speichern, mit den Fehlererfassungssymbolen, die den Daten vor einem Codieren der Symbole mittels des Fehlerkorrekturcodes zugeordnet sind, aufweist.

12. Verfahren zum Übertragen von Daten nach Anspruch 11, bei dem das Verfahren weiterhin den Schritt (150) des Entfernens der Adresse der Speichermediumeinheit, die bezeichnet ist, um die Daten zu speichern, aus den Fehlererfassungssymbolen, die Daten zugeordnet sind, die aus dem Speichermedium wiedergewonnen sind, nachdem die Symbole mittels des Fehlerkorrekturcodes decodiert sind, jedoch bevor die Fehlererfassungssymbole verwendet werden, um Fehler in den wiedergewonnenen Daten zu erfassen, aufweist.

Revendications

1. Système de transmission de données pour transmettre des données entre un bus de données (9) qui transmet des données sous forme de paquets d'octets multiples comprenant une pluralité de bits de données et de bits de parité associés et un symbole de somme pour contrôle du paquet, et un dispositif de stockage des données (18), le système comprenant :

A. Un microprocesseur (12) pour contrôler une opération de transfert des données;

B. Une interface de bus (10) qui comprend :

i. des moyens de réception (20) pour recevoir les données venant du bus de données,

ii. des moyens de contrôle de parité des octets (22) pour détecter les erreurs dans les octets de données reçus en utilisant les bits de parité reçus.

iii. des moyens convertisseurs (26) pour convertir les octets de données en mots de données et associer à chaque mot de données un nombreprédéterminé de bits de parité,

iv. des moyens de contrôle de parité des mots (30) pour détecter les erreurs dans les mots de données en utilisant les bits de parité associés,

v. des moyens de codage/décodage (28) pour coder un nombre prédéterminé de mots de données et créer des symboles de détection d'erreurs.

vi. des moyens de détection d'erreurs de la somme de contrôle (24) pour détecter les erreurs dans un paquet reçu d'octets multiples en utilisant le symbole de somme de contrôle du paquet reçu,

   l'interface de bus (10) comprenant des moyens pour affirmer un signal d'erreur, interrompre le microprocesseur (12) et interrompre une opération de transmission de données toutes les fois qu'elle détecte une erreur.

C. des moyens de mémoire tampon (16) pour recevoir à partir de l'interface de bus des données en groupes contenant un nombre prédéterminé de mots de données et de symboles de détection d'erreurs associés et stocker chaque groupe dans une mémoire tampon désignée.

D. une interface de mémoire (14) pour extraire les groupes de données un par un à partir des moyens de mémoire tampon (16) et coder les données extraites en utilisant un code de correction d'erreurs pour créer des symboles de correction d'erreurs.

E. un dispositif de stockage (18) pour recevoir les données et les symboles de correction d'erreurs venant de l'interface de mémoire (14) et stocker les données et les symboles de correction d'erreur dans une unité de stockage sur un support de données.

2. Système de transmission de données selon la revendication 1, dans lequel les moyens de mémoire tampon (16) comprennent une pluralité de mémoires tampons (B1, B2), qui ont la taille d'une unité de stockage de données associées aux supports de données, ces mémoires tampons étant reliées par des indicateurs (Pi) qui pointent d'une mémoire tampon à l'autre dans l'ordre où les mémoires tampons sont remplies.

3. Système de transmission de données selon la revendication 2, dans lequel les moyens de codage/décodage (28) codent le nombre de mots de données nécessaires pour une unité de stockage des données pour créer les symboles de détection d'erreurs.

4. Système de transmission de données selon la revendication 3, dans lequel l'interface de bus (10) comprend en plus des moyens pour produire des octets zéro pour remplir à zéro une mémoire tampon (B1, B2), si les données reçues ne remplissent pas complètement chacune des mémoires tampons associées au paquet reçu, l'interface de bus produisant des bits de parité associés et comprenant des octets zéro dans les calculs de symboles de détection d'erreurs.

5. Système de transmission de données selon la revendication 4, dans lequel le microprocesseur (12) code les symboles de détection d'erreurs dans une mémoire tampon en combinant les symboles de détection d'erreurs aux symboles associés à l'adresse d'une unité de stockage de données désignée pour conserver les données, avant que les symboles de détection d'erreurs ne soient codés en utilisant un code de correction d'erreurs.

6. Système de transmission de données selon la revendication 5, dans lequel le système comprend de plus :

A. dans les moyens de la mémoire tampon (16), des moyens pour recevoir des données venant de l'interface de mémoire et stocker les données dans des groupes de la taille d'unités de mémoire dans les mémoires tampons (B1, B2);

B. dans l'interface de bus (10)

i. dans les moyens de contrôle de parité du mot (30), des moyens (29) pour produire des bits de parité pour un mot de données extrait des moyens de la mémoire tampon (16),

ii. dans les moyens convertisseurs (26), des moyens pour convertir les mots de données reçus en octets de données et associer à chaque octet de données, un bit de parité produit par le contrôle de parité du mot, et

iii. des moyens de transmission (20) pour transmettre des données au bus de données sous forme de paquets d'octets multiples;

   l'interface de bus comprenant des moyens pour contrôler les mots de données extraits pour erreur en utilisant les symboles de détection d'erreurs extraits avant de transmettre les données sous forme de paquets d'octets multiples au bus de données.

7. Système de transmission de données selon la revendication 6, dans lequel le microprocesseur (12) enlève, à partir des symboles de détection d'erreurs situés dans une mémoire tampon, l'adresse de la zone de mémoire désignée associée aux données dans la mémoire tampon avant de créer les symboles de détection d'erreurs et les mots de données associés mis à disposition de l'interface de bus.

8. Système de transmission de données selon la revendication 7, dans lequel l'interface de bus (10) comprend aussi des moyens de relecture des données pour :

A. relire les données transmises sur le bus de données par les moyens de transmission et

B. comparer les données de relecture aux données transmises par les moyens de transmission pour déterminer si la donnée transmise contient des erreurs.

9. Méthode de transmission des données entre un bus de données (9) et des dispositifs de stockage de 10 données (18), la méthode de transmission des données comprenant les étapes suivantes :

A. recevoir à partir du bus de données (9) un paquet de données qui comprend une pluralité d'octets de données de bits de 15 parité associés ainsi qu'un symbole de somme de contrôle de paquet.

B. détecter (100) des erreurs dans les octets de données reçus en utilisant les bits de parité reçus, et affirmer (106) un signal 20 d'erreurs et interrompre une opération de transfert de données dans le cas où des erreurs seraient détectées.

C. convertir (108) les octets de données en mots de donnéees et associer à chaque mot de 25 donnéees un nombre prédéterminé de bits de parité reçus, si aucune erreur n'est détectée dans les octets de données;

D. coder (110) un nombre prédéterminé de mots de données en utilisant un code de détection 30 d'erreurs et créer des symboles de détection d'erreurs.

E. détecter (112) des erreurs dans les mots de données en utilisant les bits de parité associés et affirmer (120) un signal 35 d'erreurs et interrompre l'opération de transfert de données si des erreurs sont détectées;

F. stocker (117) le nombre prédéterminé de mots de données et de symboles de détection d'erreurs associés dans une mémoire tampon (16, B1, B2) ayant la capacité pour conserver une unité de stockage de données et relier cette mémoire tampon par des indicateurs (Pi) avec les mémoires tampons associées aux données venant du même paquet de données ou d'un paquet reçu ultérieurement si aucune erreur n'est détectée dans le nombre prédéterminé de données;

G. répéter les étapes D à F;

H. quand tous les octets de données dans un paquet de données ont été codés et mis en mémoire, détecter (126) les erreurs dans le paquet de données en utilisant le symbole de somme de contrôle du paquet et affirmer (128) un signal d'erreur et interrompre l'opération de transfert de données dans le cas de détection d'erreurs;

I. coder (134) les mots de données et les symboles de détection d'erreurs associés dans une première mémoire tampon (B1) associée au paquet de données en utilisant un code de correction d'erreurs et en créant des symboles de correction d'erreurs.

J. stocker (139) les mots de données, les symboles de détection d'erreurs associés et les symboles de correction d'erreurs dans une unité de mémoire d'un dispositif de stockage (18) associé; et

K. répéter les étapes I et J jusqu'à ce que tous les mots de données de la mémoire tampon aient été stockés dans les unités de mémoire du dispositif de stockage associé.

10. Méthode de transmission de données selon la revendication 9, dans laquelle la méthode comprend de plus les étapes suivantes:

L. extraire (140) de l'unité de mémoire du dispositif de stockage (18) de données associé, les mots de données mis en mémoire et les symboles associés de détection d'erreurs et de correction d'erreurs;

M. détecter et corriger (142) les erreurs se trouvant dans les mots de dopnnées et les symboles de détection d'erreurs en utilisant les symboles de correction d'erreurs;

N. affirmer (146) un signal d'erreur et interrompre l'opération de transfert de données si toutes les erreurs ne peuvent être corrigées;

O. Produire (152) pour chaque mot de données un nombre prédéterminé de bits de parité, dans le cas où toutes les erreurs ont été corrigées;

P. détecter (154) les erreurs se trouvant dans les mots de données en utilisant les symboles de détection d'erreurs, affirmer un signal d'erreurs et interrompre l'opération de transfert de données si des erreurs sont détectées;

Q. convertir (156) les mots de données en octets de données et associer à chaque octet de données un des bits de parité associés au mot de données correspondant dans le cas où aucune erreur n'a été détectée;

R. répéter les étapes L à Q jusqu'à ce que les données associées à un paquet de données aient été extraites;

S. créer un symbole de somme de contrôle pour le paquet de données et

T. transmettre (170) le paquet de données sur le bus de données.

11. Méthode de transmission de données selon la revendication 10, dans laquelle la méthode comprend de plus l'étape (134) permettant de combiner l'adresse de l'unité support de stockage désignée pour stocker les données avec les symboles de détection d'erreurs associés aux données avant de coder (134) les symboles en utilisant le code de correction d'erreurs.

12. Méthode de transmission de données selon la revendication 11, dans laquelle la méthode comprend de plus l'étape (150) permettant d'enlever l'adresse de l'unité support de stockage désignée pour stocker les données à partir de symboles de détection d'erreurs associés aux données extraites du support de stockage après décodage des symboles en utilisant le code de correction d'erreurs et avant d'utiliser les symboles de détection d'erreurs pour détecter les erreurs dans les données extraites.

Drawing